github/hashcat
1
Fork 0
mirror of https://github.com/hashcat/hashcat synced 2024-11-02 20:39:22 +01:00
Code Releases Activity
0f73c778d5
hashcat/OpenCL/m11900.cl
jsteube dad03e394d Fixed two major problems 
1) SIMD code for all attack-mode

Macro vector_accessible() was not refactored and missing completely.
Had to rename variables rules_cnt, combs_cnt and bfs_cnt into il_cnt which was a good thing anyway as with new SIMD code they all act in the same way.

2) SIMD code for attack-mode 0

With new SIMD code, apply_rules_vect() has to return u32 not u32x.
This has massive impact on all *_a0 kernels.

I've rewritten most of them. Deep testing using test.sh is still required.

Some kernel need more fixes:

- Some are kind of completely incompatible like m10400 but they still use old check_* includes, we should get rid of them as they are no longer neccessary as we have simd.c
- Some have a chance but require additional effort like m11500. We can use commented out "#define NEW_SIMD_CODE" to find them

This change can have negative impact on -a0 performance for device that require vectorization. That is mostly CPU devices. New GPU's are all scalar, so they wont get hurt by this.
This change also proofes that there's no way to efficiently vectorize kernel rules with new SIMD code, but it enables the addition of the rule functions like @ that we were missing for some long time. This is a TODO.
2016-02-27 17:18:54 +01:00

407 lines
13 KiB
Common Lisp
Raw Blame History

/**
* Author......: Jens Steube <jens.steube@gmail.com>
* License.....: MIT
*/
#define _PBKDF2_MD5_
#include "include/constants.h"
#include "include/kernel_vendor.h"
#define DGST_R0 0
#define DGST_R1 1
#define DGST_R2 2
#define DGST_R3 3
#include "include/kernel_functions.c"
#include "OpenCL/types_ocl.c"
#include "OpenCL/common.c"
#define COMPARE_S "OpenCL/check_single_comp4.c"
#define COMPARE_M "OpenCL/check_multi_comp4.c"
static void md5_transform (const u32 w0[4], const u32 w1[4], const u32 w2[4], const u32 w3[4], u32 digest[4])
{
u32 a = digest[0];
u32 b = digest[1];
u32 c = digest[2];
u32 d = digest[3];
u32 w0_t = w0[0];
u32 w1_t = w0[1];
u32 w2_t = w0[2];
u32 w3_t = w0[3];
u32 w4_t = w1[0];
u32 w5_t = w1[1];
u32 w6_t = w1[2];
u32 w7_t = w1[3];
u32 w8_t = w2[0];
u32 w9_t = w2[1];
u32 wa_t = w2[2];
u32 wb_t = w2[3];
u32 wc_t = w3[0];
u32 wd_t = w3[1];
u32 we_t = w3[2];
u32 wf_t = w3[3];
MD5_STEP (MD5_Fo, a, b, c, d, w0_t, MD5C00, MD5S00);
MD5_STEP (MD5_Fo, d, a, b, c, w1_t, MD5C01, MD5S01);
MD5_STEP (MD5_Fo, c, d, a, b, w2_t, MD5C02, MD5S02);
MD5_STEP (MD5_Fo, b, c, d, a, w3_t, MD5C03, MD5S03);
MD5_STEP (MD5_Fo, a, b, c, d, w4_t, MD5C04, MD5S00);
MD5_STEP (MD5_Fo, d, a, b, c, w5_t, MD5C05, MD5S01);
MD5_STEP (MD5_Fo, c, d, a, b, w6_t, MD5C06, MD5S02);
MD5_STEP (MD5_Fo, b, c, d, a, w7_t, MD5C07, MD5S03);
MD5_STEP (MD5_Fo, a, b, c, d, w8_t, MD5C08, MD5S00);
MD5_STEP (MD5_Fo, d, a, b, c, w9_t, MD5C09, MD5S01);
MD5_STEP (MD5_Fo, c, d, a, b, wa_t, MD5C0a, MD5S02);
MD5_STEP (MD5_Fo, b, c, d, a, wb_t, MD5C0b, MD5S03);
MD5_STEP (MD5_Fo, a, b, c, d, wc_t, MD5C0c, MD5S00);
MD5_STEP (MD5_Fo, d, a, b, c, wd_t, MD5C0d, MD5S01);
MD5_STEP (MD5_Fo, c, d, a, b, we_t, MD5C0e, MD5S02);
MD5_STEP (MD5_Fo, b, c, d, a, wf_t, MD5C0f, MD5S03);
MD5_STEP (MD5_Go, a, b, c, d, w1_t, MD5C10, MD5S10);
MD5_STEP (MD5_Go, d, a, b, c, w6_t, MD5C11, MD5S11);
MD5_STEP (MD5_Go, c, d, a, b, wb_t, MD5C12, MD5S12);
MD5_STEP (MD5_Go, b, c, d, a, w0_t, MD5C13, MD5S13);
MD5_STEP (MD5_Go, a, b, c, d, w5_t, MD5C14, MD5S10);
MD5_STEP (MD5_Go, d, a, b, c, wa_t, MD5C15, MD5S11);
MD5_STEP (MD5_Go, c, d, a, b, wf_t, MD5C16, MD5S12);
MD5_STEP (MD5_Go, b, c, d, a, w4_t, MD5C17, MD5S13);
MD5_STEP (MD5_Go, a, b, c, d, w9_t, MD5C18, MD5S10);
MD5_STEP (MD5_Go, d, a, b, c, we_t, MD5C19, MD5S11);
MD5_STEP (MD5_Go, c, d, a, b, w3_t, MD5C1a, MD5S12);
MD5_STEP (MD5_Go, b, c, d, a, w8_t, MD5C1b, MD5S13);
MD5_STEP (MD5_Go, a, b, c, d, wd_t, MD5C1c, MD5S10);
MD5_STEP (MD5_Go, d, a, b, c, w2_t, MD5C1d, MD5S11);
MD5_STEP (MD5_Go, c, d, a, b, w7_t, MD5C1e, MD5S12);
MD5_STEP (MD5_Go, b, c, d, a, wc_t, MD5C1f, MD5S13);
MD5_STEP (MD5_H , a, b, c, d, w5_t, MD5C20, MD5S20);
MD5_STEP (MD5_H , d, a, b, c, w8_t, MD5C21, MD5S21);
MD5_STEP (MD5_H , c, d, a, b, wb_t, MD5C22, MD5S22);
MD5_STEP (MD5_H , b, c, d, a, we_t, MD5C23, MD5S23);
MD5_STEP (MD5_H , a, b, c, d, w1_t, MD5C24, MD5S20);
MD5_STEP (MD5_H , d, a, b, c, w4_t, MD5C25, MD5S21);
MD5_STEP (MD5_H , c, d, a, b, w7_t, MD5C26, MD5S22);
MD5_STEP (MD5_H , b, c, d, a, wa_t, MD5C27, MD5S23);
MD5_STEP (MD5_H , a, b, c, d, wd_t, MD5C28, MD5S20);
MD5_STEP (MD5_H , d, a, b, c, w0_t, MD5C29, MD5S21);
MD5_STEP (MD5_H , c, d, a, b, w3_t, MD5C2a, MD5S22);
MD5_STEP (MD5_H , b, c, d, a, w6_t, MD5C2b, MD5S23);
MD5_STEP (MD5_H , a, b, c, d, w9_t, MD5C2c, MD5S20);
MD5_STEP (MD5_H , d, a, b, c, wc_t, MD5C2d, MD5S21);
MD5_STEP (MD5_H , c, d, a, b, wf_t, MD5C2e, MD5S22);
MD5_STEP (MD5_H , b, c, d, a, w2_t, MD5C2f, MD5S23);
MD5_STEP (MD5_I , a, b, c, d, w0_t, MD5C30, MD5S30);
MD5_STEP (MD5_I , d, a, b, c, w7_t, MD5C31, MD5S31);
MD5_STEP (MD5_I , c, d, a, b, we_t, MD5C32, MD5S32);
MD5_STEP (MD5_I , b, c, d, a, w5_t, MD5C33, MD5S33);
MD5_STEP (MD5_I , a, b, c, d, wc_t, MD5C34, MD5S30);
MD5_STEP (MD5_I , d, a, b, c, w3_t, MD5C35, MD5S31);
MD5_STEP (MD5_I , c, d, a, b, wa_t, MD5C36, MD5S32);
MD5_STEP (MD5_I , b, c, d, a, w1_t, MD5C37, MD5S33);
MD5_STEP (MD5_I , a, b, c, d, w8_t, MD5C38, MD5S30);
MD5_STEP (MD5_I , d, a, b, c, wf_t, MD5C39, MD5S31);
MD5_STEP (MD5_I , c, d, a, b, w6_t, MD5C3a, MD5S32);
MD5_STEP (MD5_I , b, c, d, a, wd_t, MD5C3b, MD5S33);
MD5_STEP (MD5_I , a, b, c, d, w4_t, MD5C3c, MD5S30);
MD5_STEP (MD5_I , d, a, b, c, wb_t, MD5C3d, MD5S31);
MD5_STEP (MD5_I , c, d, a, b, w2_t, MD5C3e, MD5S32);
MD5_STEP (MD5_I , b, c, d, a, w9_t, MD5C3f, MD5S33);
digest[0] += a;
digest[1] += b;
digest[2] += c;
digest[3] += d;
}
static void hmac_md5_pad (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], u32 ipad[4], u32 opad[4])
{
w0[0] = w0[0] ^ 0x36363636;
w0[1] = w0[1] ^ 0x36363636;
w0[2] = w0[2] ^ 0x36363636;
w0[3] = w0[3] ^ 0x36363636;
w1[0] = w1[0] ^ 0x36363636;
w1[1] = w1[1] ^ 0x36363636;
w1[2] = w1[2] ^ 0x36363636;
w1[3] = w1[3] ^ 0x36363636;
w2[0] = w2[0] ^ 0x36363636;
w2[1] = w2[1] ^ 0x36363636;
w2[2] = w2[2] ^ 0x36363636;
w2[3] = w2[3] ^ 0x36363636;
w3[0] = w3[0] ^ 0x36363636;
w3[1] = w3[1] ^ 0x36363636;
w3[2] = w3[2] ^ 0x36363636;
w3[3] = w3[3] ^ 0x36363636;
ipad[0] = MD5M_A;
ipad[1] = MD5M_B;
ipad[2] = MD5M_C;
ipad[3] = MD5M_D;
md5_transform (w0, w1, w2, w3, ipad);
w0[0] = w0[0] ^ 0x6a6a6a6a;
w0[1] = w0[1] ^ 0x6a6a6a6a;
w0[2] = w0[2] ^ 0x6a6a6a6a;
w0[3] = w0[3] ^ 0x6a6a6a6a;
w1[0] = w1[0] ^ 0x6a6a6a6a;
w1[1] = w1[1] ^ 0x6a6a6a6a;
w1[2] = w1[2] ^ 0x6a6a6a6a;
w1[3] = w1[3] ^ 0x6a6a6a6a;
w2[0] = w2[0] ^ 0x6a6a6a6a;
w2[1] = w2[1] ^ 0x6a6a6a6a;
w2[2] = w2[2] ^ 0x6a6a6a6a;
w2[3] = w2[3] ^ 0x6a6a6a6a;
w3[0] = w3[0] ^ 0x6a6a6a6a;
w3[1] = w3[1] ^ 0x6a6a6a6a;
w3[2] = w3[2] ^ 0x6a6a6a6a;
w3[3] = w3[3] ^ 0x6a6a6a6a;
opad[0] = MD5M_A;
opad[1] = MD5M_B;
opad[2] = MD5M_C;
opad[3] = MD5M_D;
md5_transform (w0, w1, w2, w3, opad);
}
static void hmac_md5_run (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], u32 ipad[4], u32 opad[4], u32 digest[4])
{
digest[0] = ipad[0];
digest[1] = ipad[1];
digest[2] = ipad[2];
digest[3] = ipad[3];
md5_transform (w0, w1, w2, w3, digest);
w0[0] = digest[0];
w0[1] = digest[1];
w0[2] = digest[2];
w0[3] = digest[3];
w1[0] = 0x80;
w1[1] = 0;
w1[2] = 0;
w1[3] = 0;
w2[0] = 0;
w2[1] = 0;
w2[2] = 0;
w2[3] = 0;
w3[0] = 0;
w3[1] = 0;
w3[2] = (64 + 16) * 8;
w3[3] = 0;
digest[0] = opad[0];
digest[1] = opad[1];
digest[2] = opad[2];
digest[3] = opad[3];
md5_transform (w0, w1, w2, w3, digest);
}
__kernel void m11900_init (__global pw_t *pws, __global kernel_rule_t *rules_buf, __global comb_t *combs_buf, __global bf_t *bfs_buf, __global pbkdf2_md5_tmp_t *tmps, __global void *hooks, __global u32 *bitmaps_buf_s1_a, __global u32 *bitmaps_buf_s1_b, __global u32 *bitmaps_buf_s1_c, __global u32 *bitmaps_buf_s1_d, __global u32 *bitmaps_buf_s2_a, __global u32 *bitmaps_buf_s2_b, __global u32 *bitmaps_buf_s2_c, __global u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global digest_t *digests_buf, __global u32 *hashes_shown, __global salt_t *salt_bufs, __global pbkdf2_md5_t *esalt_bufs, __global u32 *d_return_buf, __global u32 *d_scryptV_buf, const u32 bitmap_mask, const u32 bitmap_shift1, const u32 bitmap_shift2, const u32 salt_pos, const u32 loop_pos, const u32 loop_cnt, const u32 il_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max)
{
/**
* base
*/
const u32 gid = get_global_id (0);
if (gid >= gid_max) return;
u32 w0[4];
w0[0] = pws[gid].i[ 0];
w0[1] = pws[gid].i[ 1];
w0[2] = pws[gid].i[ 2];
w0[3] = pws[gid].i[ 3];
u32 w1[4];
w1[0] = pws[gid].i[ 4];
w1[1] = pws[gid].i[ 5];
w1[2] = pws[gid].i[ 6];
w1[3] = pws[gid].i[ 7];
u32 w2[4];
w2[0] = pws[gid].i[ 8];
w2[1] = pws[gid].i[ 9];
w2[2] = pws[gid].i[10];
w2[3] = pws[gid].i[11];
u32 w3[4];
w3[0] = pws[gid].i[12];
w3[1] = pws[gid].i[13];
w3[2] = pws[gid].i[14];
w3[3] = pws[gid].i[15];
/**
* salt
*/
const u32 salt_len = salt_bufs[salt_pos].salt_len;
u32 esalt_buf0[4];
u32 esalt_buf1[4];
u32 esalt_buf2[4];
u32 esalt_buf3[4];
esalt_buf0[0] = esalt_bufs[salt_pos].salt_buf[ 0];
esalt_buf0[1] = esalt_bufs[salt_pos].salt_buf[ 1];
esalt_buf0[2] = esalt_bufs[salt_pos].salt_buf[ 2];
esalt_buf0[3] = esalt_bufs[salt_pos].salt_buf[ 3];
esalt_buf1[0] = esalt_bufs[salt_pos].salt_buf[ 4];
esalt_buf1[1] = esalt_bufs[salt_pos].salt_buf[ 5];
esalt_buf1[2] = esalt_bufs[salt_pos].salt_buf[ 6];
esalt_buf1[3] = esalt_bufs[salt_pos].salt_buf[ 7];
esalt_buf2[0] = esalt_bufs[salt_pos].salt_buf[ 8];
esalt_buf2[1] = esalt_bufs[salt_pos].salt_buf[ 9];
esalt_buf2[2] = esalt_bufs[salt_pos].salt_buf[10];
esalt_buf2[3] = esalt_bufs[salt_pos].salt_buf[11];
esalt_buf3[0] = esalt_bufs[salt_pos].salt_buf[12];
esalt_buf3[1] = esalt_bufs[salt_pos].salt_buf[13];
esalt_buf3[2] = (64 + salt_len + 4) * 8;
esalt_buf3[3] = 0;
u32 ipad[4];
u32 opad[4];
hmac_md5_pad (w0, w1, w2, w3, ipad, opad);
tmps[gid].ipad[0] = ipad[0];
tmps[gid].ipad[1] = ipad[1];
tmps[gid].ipad[2] = ipad[2];
tmps[gid].ipad[3] = ipad[3];
tmps[gid].opad[0] = opad[0];
tmps[gid].opad[1] = opad[1];
tmps[gid].opad[2] = opad[2];
tmps[gid].opad[3] = opad[3];
for (u32 i = 0, j = 1; i < 4; i += 4, j += 1)
{
u32 dgst[4];
hmac_md5_run (esalt_buf0, esalt_buf1, esalt_buf2, esalt_buf3, ipad, opad, dgst);
tmps[gid].dgst[i + 0] = dgst[0];
tmps[gid].dgst[i + 1] = dgst[1];
tmps[gid].dgst[i + 2] = dgst[2];
tmps[gid].dgst[i + 3] = dgst[3];
tmps[gid].out[i + 0] = dgst[0];
tmps[gid].out[i + 1] = dgst[1];
tmps[gid].out[i + 2] = dgst[2];
tmps[gid].out[i + 3] = dgst[3];
}
}
__kernel void m11900_loop (__global pw_t *pws, __global kernel_rule_t *rules_buf, __global comb_t *combs_buf, __global bf_t *bfs_buf, __global pbkdf2_md5_tmp_t *tmps, __global void *hooks, __global u32 *bitmaps_buf_s1_a, __global u32 *bitmaps_buf_s1_b, __global u32 *bitmaps_buf_s1_c, __global u32 *bitmaps_buf_s1_d, __global u32 *bitmaps_buf_s2_a, __global u32 *bitmaps_buf_s2_b, __global u32 *bitmaps_buf_s2_c, __global u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global digest_t *digests_buf, __global u32 *hashes_shown, __global salt_t *salt_bufs, __global pbkdf2_md5_t *esalt_bufs, __global u32 *d_return_buf, __global u32 *d_scryptV_buf, const u32 bitmap_mask, const u32 bitmap_shift1, const u32 bitmap_shift2, const u32 salt_pos, const u32 loop_pos, const u32 loop_cnt, const u32 il_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max)
{
const u32 gid = get_global_id (0);
if (gid >= gid_max) return;
u32 ipad[4];
ipad[0] = tmps[gid].ipad[0];
ipad[1] = tmps[gid].ipad[1];
ipad[2] = tmps[gid].ipad[2];
ipad[3] = tmps[gid].ipad[3];
u32 opad[4];
opad[0] = tmps[gid].opad[0];
opad[1] = tmps[gid].opad[1];
opad[2] = tmps[gid].opad[2];
opad[3] = tmps[gid].opad[3];
for (u32 i = 0; i < 4; i += 4)
{
u32 dgst[4];
dgst[0] = tmps[gid].dgst[i + 0];
dgst[1] = tmps[gid].dgst[i + 1];
dgst[2] = tmps[gid].dgst[i + 2];
dgst[3] = tmps[gid].dgst[i + 3];
u32 out[4];
out[0] = tmps[gid].out[i + 0];
out[1] = tmps[gid].out[i + 1];
out[2] = tmps[gid].out[i + 2];
out[3] = tmps[gid].out[i + 3];
for (u32 j = 0; j < loop_cnt; j++)
{
u32 w0[4];
u32 w1[4];
u32 w2[4];
u32 w3[4];
w0[0] = dgst[0];
w0[1] = dgst[1];
w0[2] = dgst[2];
w0[3] = dgst[3];
w1[0] = 0x80;
w1[1] = 0;
w1[2] = 0;
w1[3] = 0;
w2[0] = 0;
w2[1] = 0;
w2[2] = 0;
w2[3] = 0;
w3[0] = 0;
w3[1] = 0;
w3[2] = (64 + 16) * 8;
w3[3] = 0;
hmac_md5_run (w0, w1, w2, w3, ipad, opad, dgst);
out[0] ^= dgst[0];
out[1] ^= dgst[1];
out[2] ^= dgst[2];
out[3] ^= dgst[3];
}
tmps[gid].dgst[i + 0] = dgst[0];
tmps[gid].dgst[i + 1] = dgst[1];
tmps[gid].dgst[i + 2] = dgst[2];
tmps[gid].dgst[i + 3] = dgst[3];
tmps[gid].out[i + 0] = out[0];
tmps[gid].out[i + 1] = out[1];
tmps[gid].out[i + 2] = out[2];
tmps[gid].out[i + 3] = out[3];
}
}
__kernel void m11900_comp (__global pw_t *pws, __global kernel_rule_t *rules_buf, __global comb_t *combs_buf, __global bf_t *bfs_buf, __global pbkdf2_md5_tmp_t *tmps, __global void *hooks, __global u32 *bitmaps_buf_s1_a, __global u32 *bitmaps_buf_s1_b, __global u32 *bitmaps_buf_s1_c, __global u32 *bitmaps_buf_s1_d, __global u32 *bitmaps_buf_s2_a, __global u32 *bitmaps_buf_s2_b, __global u32 *bitmaps_buf_s2_c, __global u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global digest_t *digests_buf, __global u32 *hashes_shown, __global salt_t *salt_bufs, __global pbkdf2_md5_t *esalt_bufs, __global u32 *d_return_buf, __global u32 *d_scryptV_buf, const u32 bitmap_mask, const u32 bitmap_shift1, const u32 bitmap_shift2, const u32 salt_pos, const u32 loop_pos, const u32 loop_cnt, const u32 il_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max)
{
/**
* base
*/
const u32 gid = get_global_id (0);
if (gid >= gid_max) return;
const u32 lid = get_local_id (0);
const u32 r0 = tmps[gid].out[DGST_R0];
const u32 r1 = tmps[gid].out[DGST_R1];
const u32 r2 = tmps[gid].out[DGST_R2];
const u32 r3 = tmps[gid].out[DGST_R3];
#define il_pos 0
#include COMPARE_M
}
View Git Blame Copy Permalink